J . Am. Chem. SOC.1993, 115, 11581-11589
11581
Identification and Removal of Impediments to Biocatalytic Synthesis of Aromatics from D-Glucose: Rate-Limiting Enzymes in the Common Pathway of Aromatic Amino Acid Biosynthesis K. A. Dell and J. W. Frost' Contributionfrom the Department of Chemistry, Purdue University, West Lafayette, Indiana 47907 Received July 26, 1993'
Abstract: Increasing the flow of D-glucose equivalents into the common pathway of aromatic amino acid biosynthesis creates a metabolic situation where individual common pathway enzymes become rate-limiting. Such enzymes are unable to convert substrate to product at a rate sufficient to avoid intracellular accumulation of substrate. Export of the accumulating substrate by the host microbe into its culture supernatant results in lowered percent conversions and decreased purity of desired aromatic products. Methodology has now been developed which facilitates rapid identification and removal of impediments to biocatalytic synthesis of aromatics from D-glucose which are caused by rate-limiting, common pathway enzymes. An Escherichia coli mutant, D2704, incapable of L-tryptophan and L-tyrosine synthesis was transformed with a plasmid which increased the in vivo catalytic activity of 3-deoxy-~-arabino-heptulosonic acid 7-phosphate (DAHP) synthase and transketolase. Analysis by lH N M R of the culture supernatant of this construct revealed the accumulation of L-phenylalanine and phenyllactate along with common pathway enzyme substrates and related metabolites. Plasmid-borne genes encoding common pathway enzymes were then introduced into D2704 individually and in various combinations followed by 'H N M R analysis of these constructs' culture supernatants. Progress toward increasing the catalytic activity of rate-limiting enzymes was indicated by a decrease in the number of accumulated enzyme substrates and increased L-phenylalanine synthesis and phenyllactate synthesis. In this fashion, 3-dehydroquinate (DHQ) synthase, shikimate kinase, 5-enolpyruvoylshikimate 3-phosphate (EPSP) synthase, and chorismate synthase were identified as rate-limiting enzymes. A feedback loop in E. coli was also identified involving inhibition of shikimate dehydrogenase by shikimic acid. Although microbial aromatic amino acid biosynthesis' can be manipulated to produce an impressive array of aromatic structures from nontoxic, inexpensive D-glucose, several different challenges confront the use of intact microbes as catalysts in the synthesis of industrial and medicinal aromatics.* Those factors which affect yield, conversion rate, or product purity are of particular importance. For instance, the highest possible percentage of D-glUcOSeCO"!d by themicrobemust bedirected into synthesis of 3-deoxy-~-arabino-heptulosonic acid 7-phosphate (DAHP), the first committed intermediate of aromatic amino acid biosynthesis (Scheme I). This surge in the synthesis of DAHP has to then bedelivered by way of the enzymes of the common pathway of aromatic amino acid biosynthesis (Scheme I) into increased synthesis of chorismic acid. Finally, carbon flow at the end of the common pathway needs to be directed into the various pathways which convert chorismate into L-phenylalanine, L-tyrosine, and L-tryptophan along with a host of related secondary metabolites. Prior efforts have successfully identified methods for increasing the percentage of D-glucose consumed by Escherichia coli which is channeled into the common pathway. Such methods include Abstract published in Advance ACS Abstracts, November 1, 1993. (1) For an abbreviated list of relevant reviews, see: (a) Campbell, M. M.; Sainsbury, M.; Searle, P. A. Synthesis 1993, 179. (b) Dewick, P. M. Not. Prod. Rep. 1992,9,153. (c) Bentley, R. Crit. Rev. Eiochem. Mol. Eiol. 1990, 25,307. (d) Pittard, A. J. In Escherichia coli andSalmonella typhimurium; Neidhardt, F.C., Ed.; American Society for Microbiology: Washington, DC, 1987; Vol. 1, p 368. ( e ) The Shikimate Pathway; Conn, E. E., Ed.; Plenum: New York, 1986. (f) Herrmann, K.M. In Amino Acids: Biosynthesis and Genetic Regulation; Herrmann, K. M., Somerville, R. L., Ed.; AddisonWesley: Reading, 1983: p301. (8) Weiss,U.;Edwards, J. M. TheEiosynthesis of Aromatic Compounds; Wiley: New York, 1980. (h) Haslam, E. The Shikimate Pathway; Wiley: New York, 1974. (2) (a) Frost, J. W.; Lievense, J. New J . Chem., in press. (b) Draths, K. M.; Ward, T. L.; Frost, J. W. J. Am. Chem. SOC.1992, 114, 9725.
increasing the in vivo catalytic activity of DAHP ~ y n t h a s eand, ~.~ more recently, amplified expression of transketolase.5 Delivery of the surge in DAHP synthesis into increased synthesis of chorismic acid, which is the focus of this study, has received much less attention.6 Increases in the in vivo catalytic activity of DAHP synthase and transketolase create a metabolic situation where individual common pathway enzymes become rate-limiting. Such enzymes are unable to catalyze conversion of substrate to product at a rate sufficient to avoid intracellular accumulation of substrate. Because of the rapid export by E. coli of intracellularly accumulating metabolites into the culture supematant,68v7substrates of rate-limiting enzymes are effectively lost to aromatic biosynthesis. Resulting reduction in percent yield, conversion rate, and decreased purity of aromatic product are major problems. The ability of microbes to export accumulating intermediate metabolites has now been employed to biocatalytic advantage as the basis for a method of identifying rate-limiting enzymes in the (3) (a) Cobbett, C. S.;Morrison, S.;Pittard, J. J. Eacteriol. 1984, 157, 303. (b) Garner, C. C.; Herrmann, K. M. J. Eiol. Chem. 1985, 260, 3820. (c) Cobbett, C. S.;Delbridge, M. L. J. Eacteriol. 1987, 169, 2500. (4) (a) Ogino, T.; Garner, C.; Markley, J. L.; Herrmann, K. M. Proc. Narl. Acad. Sei. U.S.A. 1982, 79, 5828. (b) Weaver, L. M.; Herrmann, K.M. J. Eacteriol. 1990, 172, 6581. (5) (a) Draths, K. M.; Frost, J. W. J. Am. Chem. Soc. 1990,112, 1657. (b) Draths, K.M.; Pompliano, D. L.; Conley, D. L.; Frost, J. W.; Berry, A.; Disbrow, G. L.; Staversky, R. J.; Lievense, J. J. Am. Chem. Soc. 1992.114,
3956. (c) Frost, J. W. U S . Patent 5 168 056,1992. (d) Draths, K. M. Ph.D. Thesis, Stanford University, July 1991. (6) (a) Draths, K. M.; Frost, J. W. J. Am. Chem. Soc. 1990, 112, 9630. (b) Matsui, K.; Miwa, K.; Sano, K. U S . Patent 5 017 481, 1991. (c) Tribe, D. E. US.Patent 4 681 852, 1987. (7) (a) Frost, J. W.; Knowles, J. R. Biochemistry 1984, 23, 4465. (b) Fischer, R. S.;Berry, A.; Gaines, C. G.; Jensen, R. A. J. Eacferiol. 1986,168, 1147. (c) Draths, K. M.; Frost, J. W. J. Am. Chem. SOC.1991, 113, 9361. (d) Murdock, D.; Ensley, B. D.; Serdar, C.; Thalen, M. Biotechnology 1993, 1 1 , 381.
OOO2-7863/93/1515-11581$04.00/00 1993 American Chemical Society
Dell and Frost
11582 J. Am. Chem. SOC., Vol. 115, No. 24, 1993 A
Scheme Io
PEP
E4P
lA
5:
RO
6H
R = PO3H2 DAHP R-H DAH
le OH
1""
GI DAH Conc.
6": -tJ 6" - Q
DHS
2
R Shik s-3-P
1
-
L phenylalanine
n
"
1 2
3
4
5
6 7
8
Strain
9 1011
COzH
phenylpyruvic acid
1
16 I
B
phenyllacticacid
12 _10 _
Conc. (mM)
'
m
II 24hr
8
El 48hr
6 DHS
prephenic acid
-L - tyrosine
4
2
n 1 2 3 4 5 6 7 8 9 1 0 1 1 OH shikimic acid
6H chorismic acid
tG S3P
-L -tryptophan
EPSP
O(A) DAHP synthase (aroF aroG aroH); (B) DHQ synthase (aroB); (C) DHQ dehydratase (aroD);(D) shikimate dehydrogenase (aroE); (E) shikimate kinase (aroL aroK); (F) EPSP synthase (aroA); (G) chorismate synthase (aroC); (H) chorismate mutase (pheA tyrA); (I) prephenate dehydratase (pheA);(J) aromatic aminotransferase (tyrB).
common pathway of aromatic amino acid biosynthesis. 'H N M R is used to identify common pathway substrates and related metabolites which accumulate in the culture supernatant of an E . coli auxotroph constructed to express amplified levelsof DAHP synthase and transketolase. Genes encoding common pathway enzymes whose substrates or related metabolites accumulate are then introduced into the E. coli auxotroph. The disappearance of a previously accumulating metabolite is interpreted as evidence for an increase in the catalytic activity of an enzyme to a point where it is no longer rate-limiting. In this fashion, constructs have been assembled in which increases in carbon flow can be delivered from the beginning to the end of the common pathway of aromatic amino acid biosynthesis. Results Analysis for Rate-Limiting Enzymes. Identification and removal of impediments to the flow of carbon through the common pathway followed from the analysis by lH N M R of the culture supernatant of E . coli D2704,* a pheA, tyrA, AtrpE-C strain requiring L-tyrosine and L-tryptophan supplementation for growth. The various mutations in D2704 ensure that the chorismate produced by the common pathway is converted into phenylpyruvate, which subsequently partitions between L-phenylalanine and phenyllactic acid biosynthesis. A surge of carbon flow through thecommon pathway of D2704 was created by transforming this (8) Mascarenhas,D.; Ashworth, D. J.; Chen, C. S.Appl. Environ. Microbiol. 1991,57, 2995.
Strain
Figure 1. (A) Common pathway intermediates extracellularly accumulated by D2704 strains after 24 h of culturing in minimal medium which initially contained 56 mM D-glucose. (B) Average, combined accumulations of L-phenylalanineand phenyllactate after 24 h and 48 h. Strains studied include (1) D2704/pKD130A, (2) D2704/pKD136; (3) D2704/ pKD136/pKD28; (4) D2704/pKD136/pKAD34; ( 5 ) D2704/ pKD136/pKAD31; (6) D2704/pKD136/pKAD38; (7) D2704/pKD136/ pKAD43; (8) D2704/pKD136/pKAD39; (9) D2704/pKD136/pKAD5 1; (10) D2704/pKD136/pKAD44; (1 1) D2704/pKD136/pKAD50.
auxotroph with P K D ~ ~ O A a plasmid , ~ ~ which carried the tkt gene encoding transketolase and the aroF gene encoding the tyrosine-sensitive isozyme of DAHP synthase. As with all the constructs examined, D2704/pKD 130A was initially grown in rich medium (LB), harvested, and then cultured for 48 h in minimal medium (M9) containing 56 mM D-glucose. A portion of the culture supernatant was withdrawn from the minimal medium at 24 and 48 h and analyzed by lH NMR. Accumulation of common pathway metabolites was best observed a t 24 h of culturing in minimal medium while end-product L-phenylalanine and phenyllactate concentrations reached their maximum values by 48 h of culturing in minimal medium. 3-Deoxy-~-arabinoheptulosonicacid (DAH), 3-dehydroshikimate (DHS), shikimate, and shikimate 3-phosphate (S3P) were detected after 24 h of culturing D2704/pKD130A in minimal medium (Figure 1A/ entry 1 and Figure 2A). On the basis of the enzymes for which these or closely related metabolites are substrates, DHQ synthase, shikimate dehydrogenase, shikimate kinase, and EPSP synthase were tentatively identified as rate-limiting enzymes. Phenyllactate and L-phenylalanine were detected in the culture supernatant of D2704/pKD130A after 24 h (Figure lB/entry 1, Figure 2A) and increased to 5.6 f 0.7 mM after 48 h of culturing in minimal medium (Figure 1B/entry 1). Prephenate was present in the culture supernatant of D2704/pKD130A at 24 h but had largely disappeared by 48 h of culturing in minimal medium. Preventing Accumulation of Pathway Intermediates. Previous efforts6a have shown DHQ synthase to be a metabolic block in the strain AB2834, an aroE auxotroph, which can be removed when in vivo activities of the enzyme are increased. Transformation of D2704 with pKD136, a plasmid6a which carries aroB (encoding DHQ synthase), tkt (encoding transketolase), and aroF (encoding the tyrosine-sensitive isozyme of DAHP synthase), resulted (Figure 1A, lB/entry 2) in the disappearance of DAH from the culture supernatant and increased accumulation of DHS,
Biocatalytic Synthesis of Aromatics from
D- Glucose
J . Am. Chem. SOC.,Vol. 1 1 5, No. 24, 1993 11583
A
B
Figure 2. (A) Before removal of impediments to carbon flow: 'H NMR of the supernatant of D2704/pKD130A after 24 h of culturing in minimal medium initially containing 56 mM pglucose. Observable resonances for accumulated molecules are DAH (Ha,He, Hc), DHS (HD,HE),shikimate (HF,Ho, HH),shikimate 3-phosphate (HI, HJ), prephenic acid (HK,HL. HM,HN,Ho), phenyllactic acid (Hp and aromatic resonances at 7.4 ppm), and phenylalanine (aromatic resonances at 7.4 ppm). (B) After removal of impediments to carbon flow: 'H NMR of the culture supernatant of D2704/pKD136/pKAD50 after 48 h. Observable resonances for accumulated molecules are phenyllactic acid (Hp, HQ,and aromatic resonances at 7.4 ppm) and phenylalanine (HR,Hs, and aromatic resonances at 7.4 ppm).
shikimate, and shikimate 3-phosphate. Increased accumulation of L-phenylalanine and phenyllactate was not observed. Thus the removal of the impediment to carbon flow at DHQ synthase allowed unrestricted carbon flow through the initial portion of thecommon pathway but failed to increase end-product synthesis due to the continued presence of other rate-limiting enzymes. The lack of a suitably situated polylinker for the insertion of additional genetic fragments into pKD136 led to the use of a two-plasmid system. Plasmid pKDl36 was retained as the plasmid which directed the surge of carbon flow into the common pathway and removed the impediment to this flow of carbon a t DHQ synthase. Genes encoding other rate-limiting enzymes were inserted into the polylinker region of either pSUl8 or pSU19 to form the second plasmid (Table I). Accumulation of 3.1 f 0.1 mM phenyllactate and L-phenylalanine by D2704/pKD136/ pSU18 indicated that this second plasmid (pSU18) was a significant burden on the microbe's (D2704/pKD136) metabolism. Plasmid pKD28, derived from insertion of aroE (encoding shikimate dehydrogenase) into pSU 18 (Table I), was transformed intoD2704/pKD136in aneffort toremoveaccumulationof DHS. While a decrease in the levels of DHS in the culture supernatant of D2704/pKD136/pKD28 was observed, introduction of the
second plasmid did not completely eliminate DHS accumulation (Figure 1A/entry 3). Shikimateand shikimate 3-phosphate were still present in the culture supernatant, and the total production of L-phenylalanine and phenyllactate declined to 2.1 f 0.9 mM after 48 h. Additional insights into prevention of DHS accumulation were afforded by efforts to compensate for the ratelimiting character of shikimate kinase. Plasmid pKAD34 (Table I) wasconstructed whichcarriedaroL (encoding shikimatekinase) and aroE (encoding shikimate dehydrogenase) inserts. No DHS or shikimate could be detected in the culture supernatant of D2704/pKD136/pKAD34, leaving shikimate 3-phosphate as the only accumulated common pathway intermediate (Figure 1A/ entry 4). No improvement was achieved in the levels of L-phenylalanine and phenyllactate synthesized by D2704/ pKD136/pKAD34 (Figure lB/entry 4). DHS and shikimate were also absent in the culture supernatant (Figure lA/entry 5 ) of D2704/pKD136/pKAD31. Plasmid pKAD3 1 carried only an aroL (encoding shikimate kinase) insert (Table I). The accumulation of shikimate 3-phosphate was still observed, and the total production of L-phenylalanine and phenyllactate (Figure 1B/entry 5 ) of 5.6 f 0.5 mMwasessentially unchanged relative to the case of D2704/pKD136. Removal of both DHS and shikimate accumulation with a second plasmid
11584 J. Am. Chem. SOC.,Vol. 115, No. 24, 1993
Dell and Frost
Table I. Restriction Enzyme Maps of Plasmids Used to Increase the in Vivo Catalytic Activity of Rate-Limiting, Common Pathway Enzymes#
Plasmid
pKD28 (3.9 kb)
Vector
Plasmid Map
Pcc pKAD31 (3.3 kb)
pKAD38 (4.7 kb)
psuw
K
B
X
-
P I D
psu19
H PS
B K 1
amc
psu19
amA
psu19
--
K I
G
amL 0
PS I
I
amA
KE I
I,
D
amC
G
B
PS
I
KE I
aroA
PGC pKAD51 (5.0 kb)
G
* PSB
I
PGC
pKAD50 (7.4 kb)
G
amA
PGC pKAD44 (6.4 kb)
K
EK
psu19
psu19
CZ
1 1 8
PGC pKAD43 (5.7 kb)
-.
amL
Plac pKAD39 (4 kb)
G I
PGC psu19
-
amE
H PSXB
psUl9
pKAD34 (4.9 kb)
H
E0
psurs
0
S
>-
amC
G
amL
KE
I
~.
Restriction enzyme sites are abbreviated as follows: E = EcoRI, B = BomHI, H = HindIII, X = XbaI, K = KpnI, P = PsfI, S = SafI. possessing only an aroL insert suggested that the apparent ratelimiting character of shikimate dehydrogenase might not be due to inadequate expression levels. Rather, shikimate accumulating due to rate-limiting shikimate kinase activity might be feedback inhibiting shikimate dehydrogenase. To test this hypothesis, shikimate dehydrogenase was purified to homogeneity and the inhibition of the enzyme was measured in the presence of increasing concentrations of shikimate. Shikimate was found to be a product inhibitor exhibiting linear mixed-type inhibition9 (Figure 3) with an inhibition constant (Ki)of 0.16 mM. Improving End Product Accumulation. Removing those impediments to carbon flow responsible for DAH, DHS, and shikimate accumulation failed to provide an increase in the accumulation of L-phenylalanine and phenyllactate. This suggestion of a bottleneck later in the common pathway led to consideration of EPSP synthase, since shikimate 3-phosphate, the only accumulating common pathway metabolite which remained, was the substrate for this enzyme. Plasmid pKAD38 containing aroA (encoding EPSP synthase) was constructed (Table I) to increase the levels of EPSP synthase. Evaluation of D2704/pKD136/pKAD38 revealed total L-phenylalanine and phenyllactate production of 7.9 f 1.3 mM after 48 h (Figure lB/entry 6), which was a significant increase relative to endproduct accumulations in strains lacking plasmid-borne aroA. It is noteworthy that this improved end-product accumulation was achieved in D2704/pKD136/pKAD38 despite the continued accumulation (Figure lA/entry 6) of DHS, shikimate, and shikimate 3-phosphate. Given the earlier success in removing both DHS and shikimate accumulation with amplified shikimate (9) Segel, I. H.Biochemical Calcularions; Wiley: N e w York,-l97Fp 261.
ShildmaCe
Concentmion (mhQ
.O 8 8
'
-500 -10
I
I
I
I
J
0
10
20
30
40
m1
.13 25 .38 .5
(l/mM)
Figure 3. Lineweaver-Burk plot of shikimate dehydrogenase inhibition by product shikimate.
kinase activity, plasmid pKAD43 which carried both aroA (encoding EPSP synthase) and aroL (encoding shikimate kinase) was constructed (Table I). The strain D2704/pKD136/pKAD43 produced 9.7 f 0.3 mM of L-phenylalanine and phenyllactate (Figure lB/entry 7) with the accumulation of only one common pathway intermediate, shikimate 3-phosphate (Figure lA/entry 7). The possibility now arose that ratelimiting chorismate synthase activity might contribute to the continued presence (Figure 1A/ entry 7) of shikimate 3-phosphate in the culture supernatants of constructs such as D2704/pKD136/pKAD43 which expressed
J. Am. Chem. SOC.,Vol. 115, No. 24, 1993
Biocatalytic Synthesis of Aromatics from D-Glucose
11585
Table XI. Ratios of the Specific Activities of Amplified Common Pathway Enzymes Relative to the Specific Activities of Unamplified Common Pathway Enzymes in E. coli D2704 Constructs strain D2704‘ D2704/pKD136 D2704/pKD136/pKAD31 D2704/pKD136/pKAD38 D2704/pKD 136/pKAD43 D2704/pKD 136/pKAD50
12
DHQ shikimate EPSP chorismate synthase kinase synthase synthase 1 .o 3.1 1.2 0.7 1.1 2.0
1 .o
1 .o
10 End Product
1 .o
Accumulation (mM)
8
120 24 41
19 1.6 8.3
4 2
4.2
0
D2704 enzyme activity values (units/mg) are as follows: DHQ synthase, 0.023; shikimate kinase, 0.0023; EPSP synthase, 0.0099; chorismate synthase, 0.0017. One unit is defined as one pmol of product formed per min.
amplified EPSP synthase activity. As a result, plasmid pKAD5O was constructed (Table I) which carried aroA (encoding EPSP synthase), aroL (encoding shikimate kinase), and aroC (encoding chorismate synthase). The strain D2704/pKD136/pKAD50 accumulated small quantities of DHS and shikimate 3-phosphate (Figure lA/entry 11) at 24 hand 12.3 f 2.2 mM L-phenylalanine and phenyllactate at 48 h (Figure lB/entry 11 and Figure 2B). The observed increase in end-product levels and reduction in shikimate 3-phosphate accumulation suggested that chorismate synthase was an impediment to the flow of carbon through the common pathway. To further evaluate the role of chorismate synthase, plasmids pKAD39, pKAD5 1,and pKAD44 respectively carrying aroC,aroCaroL, and aroA aroCwere constructed (Table I) and inserted into D2704/pKD136. None of these constructs synthesized the (Figure 1B/entries 8-10) levels of end products synthesized by D2704/pKD136/pKAD50 (Figure lB/entry 11). Enzyme Activities. The ability of D2704, a pheA and tyrA mutant, to synthesize L-phenylalanine could be consistent with nonenzymatic conversion of chorismate into end product or might indicate the presence of residual chorismate mutase or prephenate dehydratase activity. Specificactivities were therefore measured in D2704 for chorismate mutase (8.5 X 10-5 units/mg) and prephenate dehydratase (2.8 X units/mg). Comparison with the specific activities of chorismate mutase (1.1 X l e 2 units/ mg) and prephenate dehydratase (9.2 X le3 units/mg) in the wild-type strain E. coli RB79 1indicatesthat enzymatic conversion of chorismate into phenylpyruvate is essentially absent in D2704. Enzyme activities were also measured (Table 11) for DHQ synthase, shikimate kinase, EPSP synthase, and chorismate synthase in those strains which either gave a significant increase in end-product accumulation or a significant improvement in product purity. The levels of overexpression (Table 11) of DHQ synthase, EPSP synthase, and chorismate synthase were generally modest relative to the overexpression of shikimate kinase. Measured Improvement in Carbon Flow. The 12.3 f 2.2 mM concentration (74% of theoretical maximum yield)h of L-phenylalanine and phenyllactate synthesized (Figure 1B/entry 11) by D2704/pKD136/pKAD50 relative to the 5.6 f 0.7 mM concentration (34% of theoretical maximum yield)2a of Lphenylalanine and phenyllactate synthesized by D2704/pKD 130A (Figure 1B/entry 1) clearly attests to the improvementin carbon flow delivered to the end of the common pathway attendant with increasing the in vivo catalytic activitiesof rate-limiting enzymes. A complementary way to measure progress toward removal of impediments to carbon flow is to determine whether incremental improvements in carbon flow directed into the common pathway translate into incremental improvements in carbon flow exiting the common pathway. As has already been mentioned, amplified in vivo activity of DAHP synthase and amplified in vivo activity of transketolase are two strategies for increasing the flow of carbon directed into aromatic amino acid biosynthesis. Therefore, it is informative to determine the impact (Figure 4) of amplified expression levels of rate-limiting, common pathway enzymes on the levels of synthesized end products when the catalytic activity
1
D2704l
2
D2704l
3
4
D2704/ D2704/
pKD13W P ~ pKDll6B U pKD130A pKAD50 pKAD9
Figure 4. Averaged, combined accumulations (48 h) of L-phenylalanine, phenyllactate, and prephenate by various E. coli D2704 constructs after 48 h of culturing in minimal medium which initially contained 56 mM D-glucose.
of DAHP synthase is varied and, in separate experiments, when the catalytic activity of transketolase is varied. An important construct employed in these analyses was D2704/ pKAD42/pKAD50. Plasmid pKAD42 carries aroB and aroF inserts but lacks a tkt insert. Culturing of D2704/pKAD42/ pKAD5O produced large amounts of acetate and lactate. Resulting acidification of the culture medium resulted in premature death of the construct. To alleviate this problem, the pH of the accumulation medium was monitored and neutralized when necessary. Maintaining a neutral pH resulted in a pronounced accumulation of prephenic acid. This may reflect prephenate’s greater stability at neutral versus acidic pH.l0 As a result, comparisonsof end-product accumulation (Figure 4) with D2704/ pKAD42/pKAD50 required quantification of prephenate concentrations in addition to L-phenylalanine and phenyllactate concentrations. D2704/pKAD42/pKAD50 was the only strain which had to be neutralized during the 48 h culturing in minimal medium. D2704/pKD116B, which expresses amplified levels of DAHP synthase activity, synthesizes 4.5 f 0.9 mM L-phenylalanine, phenyllactate, and prephenate (Figure 4/entry 3). D2704/ pKAD42/pKAD50, which expresses amplified levels of DAHP synthase activity like D2704/pKD116B, synthesizes 8.2 f 0.8 mM L-phenylalanine, phenyllactate, and prephenate (Figure 4/entry 2). The difference between these two DAHP synthase overexpressing strains is that the in vivo catalytic activities of rate-limiting, common pathway enzymes are increased in D2704/ pKAD42/pKAD50 while the activities of these same enzymes are unchanged in D2704/pKD116B. On the basis of end product accumulations, more of the incremental improvement in carbon flow directed into the common pathway by amplified DAHP synthase activity is being delivered to the end of the pathway in D2704/pKAD42/pKAD50. Amplified transketolase activity expressed by D2704/pKD130A distinguishesthis construct from D2704/pKD116B. Both of these constructs overexpress DAHP synthase. Accumulation (Figure 4/entry 4) of L-phenylalanine, phenyllactate, and prephenate in D2704/pKD130A is 5.7 f 0.6 mM while D2704/pKD116B accumulates (Figure 4/entry 3) a 4.5 f 0.9 mM concentration of end products. This percentage increase in L-phenylalanine, phenyllactate, and prephenate accumulation is significantly less than that observed when the activitiesof the rate-limiting, common pathway enzymes are simultaneously increased with transketolase activity. As an example, D2704/pKAD42/pKAD50 expresses amplified levels of DAHP synthase and rate-limiting, common pathway enzymes while D2704/pKD 136/pKAD50 overexpresses these same enzyme activities and, in addition, overexpresses (10) Weiss, U.; Gilvarg, C.; Mingioli, E. S.; Davis, B. D. Science 1954, 119, 114.
11586 J . Am. Chem. SOC., Vol. I 15, No. 24, 1993 transketolase. D2704/pKD136/pKAD50 synthesizes 12.3 f 2.2 mM L-phenylalanine, phenyllactate, and prephenate (Figure 4/entry 1) while only an 8.2 f 0.8 mM concentration of end products (Figure 4/entry 2) is synthesized by D2704/pKAD42/ pKAD 5 0.
Discussion Removal of the impediments to the flow of carbon through the common pathway of aromatic amino acid biosynthesis required (a) direction of a surge of carbon flow into the common pathway and (b) development of analysis methodology suitable for identification of rate-limiting enzymes. For the purposes of this study, the wild-type aroF locus encoding the tyrosine-sensitive isozyme of DAHP synthase was localized on a plasmid. The native, strong promoter and the copy number of the plasmid combined to afford ample overexpression of DAHP synthase. Although not pursued in this study, additional improvements in DAHP synthase in vivo catalytic activity could likely be achieved if an aroF or aroG locus containing a point mutation rendering the gene product insensitive to feedback inhibition were employed? In vivo catalytic activity of DAHP synthase can be increased to a point beyond which additional carbon flow is not directed into the common pathway. One possible explanation for this observation is that the availability of phosphoenolpyruvate (PEP) and D-erythrose 4-phosphate (E4P), the two substrates of DAHP synthase, ultimately becomes a limiting factor. Approaches to increase the in vivo concentration of both PEP11 and E4P5 have been explored. Increasing E4P availability by increasing the in vivo catalytic activity of transketolase has proven to be particularly s u c c e s s f ~ l .At ~ levels of DAHP synthase catalytic activity where additional increases in catalytic activity do not lead to increased DAHP synthesis, amplification of transketolase expression with plasmid-encoded tkt provides an approximately twofold increase in synthesis of DAHP.5b Consequently, efforts to identify and remove impediments in the common pathway have utilized pBR325 derivativessuch aspKD130AandpKD136 whichcontain transketolase-encoding tkt and DAHP synthase-encoding aroF loci. Identification of rate-limiting enzymes entailed the use of auxotrophic E. coli mutants lacking individual common pathway enzymes. For instance, E. coli aroB/pKD130A, which lacks catalytically active DHQ synthase, accumulated 20 mM 3-deoxyD-arabino-heptulosonic acid (DAH) in its culture supernatant when cultured in medium containing 56 mM D-glucose. DAH is the dephosphorylated substrate of DHQ synthase. Subsequent introduction of transketolase and DAHP synthase-encoding pKDl30A into E. coli aroD, which lacks catalytically active DHQ dehydratase, would be expected to accumulate 20 mM 3-dehydroquinate, the substrate of DHQ dehydratase. Instead, E. coli aroDlpKD130A accumulated only 14 mM DHQ when cultured in medium containing 56 mM D-glucose. In addition to DHQ, 5 mM DAH was discovered in the culture supernatant. The reduction in accumulated substrate (or closely related metabolite) is consistent with an intervening, rate-limiting enzyme. Introduction of pKD136 which encodes DHQ synthase along with transketolase and DAHP synthase into E. coli aroD resulted in disappearance of DAH from the culture supernatant and accumulation of 19 mM DHQ. These data could be interpreted as indicative of DHQ synthase being a rate-limiting enzyme whose partial blockage of the common pathway was removed upon amplification of theenzyme. E. coliaroE/pKD136, which lacked catalytically active shikimate dehydrogenase, accumulated approximately 30 mM 3-dehydroshikimate.6a This level of DHS accumulation and lack of DHQ accumulation indicated that DHQ dehydratase was not a rate-limiting enzyme.68 (11) (a) Miller, J. E.; Backman, K. C.; OConner, M. J.; Hatch, R. T. J. Ind. Mirrobiol. 1987,2,143.(b) Backman, K. C. U.S.Patent 5 169 768,1992.
Dell and Frost Because E. coli possesses two isozymes of shikimate kinase (encoded by aroK and E. coli aroL /pKD136 failed to accumulate shikimate. In addition, E. coli aroAlpKD136 (lacking EPSP synthase) and E. coli aroClpKD136 (lacking chorismate synthase) failed to accumulate significant concentrations of substrates of the missing enzymes.7c This led to examination of E. coli D2704 (lacking chorismate mutase, prephenate dehydratase, prephenate dehydrogenase, and anthranilate synthase) which is unable to process chorismic acid due to mutations in thepheA and fyrAloci and a deletion mutation spanning trpE-C .* D2704 was initially chosen for its expected ability to accumulate chorismate. However, D2704 was found to accumulate primarily L-phenylalanine along with smaller amounts of phenyllactic acid. This reflects the precedented ability of chorismate mutase to spontaneously undergo nonen~ymaticl~ rearrangement to prephenic acid, whose transient formation by D2704 can be detected. Nonenzymatic10 decarboxylation and dehydration of prephenate affords phenylpyruvate, which undergoes predominant transamination to form L-phenylalanine while the remaining portion is reduced to phenyllactate. Although tyrB-encoded aromatic aminotransferase, aspCencoded aspartate transaminase, and ilvPencoded branchedchain aminotransferase can catalyze the transamination leading to L-phenylalanine, the aromatic aminotrahsferase has the greatest affinity for the phenylpyruvate.I5Jd That not all of the phenylpyruvate is transaminated reflects insufficient in vivo activity of the aromatic aminotransferase. This may indicate that the flow of carbon directed into phenylpyruvate synthesis is in excess of what the aromatic aminotransferase can transaminate. Alternatively, the initial culturing of the D2704 variants in rich medium may result in the microbe's consumption of enough L-tyrosine that expression of tyrR regulon genes such as the fyrB locus which encodes the aromatic transaminase may be transcriptionally repressed16 even during subsequent culturing in medium lacking L-tyrosine supplementation. Lower aromatic aminotransferase activity may also reflect disruption of the putative complexes which this transferase forms in vivo with chorismate mutase-prephenate dehydrogenase and chorismate mutase-prephenate dehydratase." This could be due to the mutations which have been introduced into the tyrA and pheA loci, which respectively encode the aforementioned bifunctional enzymes. Reduction of accumulating phenylpyruvate to phenyllactate may be exploited by the microbe as a mechanism for cycling NADH back to NAD+ when the culture medium becomes partially limiting in oxygen availability. Reduction of another a-ketocarboxylic acid, pyruvic acid, by lactate dehydrogenase is a route for regenerating NAD+ under anaerobic conditions.'* Measurement of carbon flow through the common pathway of aromatic amino acid biosynthesis is simplified in E. coli D2704, since two of the three major routes by which chorismic acid is processed have been eliminated. The combined concentration of L-phenylalanine and phenyllactate determined by 1H N M R was then a convenient way to quantitate the carbon flow delivered to the end of the common pathway. This follows from the one to one mole relationship between chorismate processed and Lphenylalanine synthesized as well as chorismate processed and phenyllactate synthesized. Determination of which enzymes in the common pathway are rate-limiting was then based on the identification of metabolites other than L-phenylalanine and (12) Lobner-Olesen, A.; Marinus, M. G. J. Bucteriol. 1992, 174, 525. (13) (a) DeFeyter, R. C.; Pittard, J. J. Bucteriol. 1986, 165, 226. (b) DeFeyter, R. C.; Davidson, B. E.; Pittard, J. J. Bucteriol. 1986, 165, 233. (14) (a) Addadi, L.; Jaffe, E. K.; Knowles,J. R. Biochemistry 1983.22, 4494. (b) Copley, S. D.; Knowles,J. R. J. Am. Chem. Soc. 1987,109,5008. (15) Tan-Wilson, A. L. In Amino Acids: Biosynthesis and Genetic
Regulation; Herrmann, K. M., Somerville, R. L., Ed.; Addison-Wesley: Reading, MA, 1983; p 19. (16) Yang, J.; Pittard, J. J. Bucteriol. 1987, 169, 4710. (17) Powell, J. T.; Morrison, J. F. Biochim. Biophys. Actu 1979,568,467. (18) Gottschalk, G. Bucteriul Metabolism; Springer-Verlag: New York, 1986.
Biocatalytic Synthesis of Aromatics from
D - Glucose
phenyllactate, which accumulated in the culture supernatant of E. coli D2704/pKD130A. In this fashion, a single 'H N M R was able to quantitate carbon flow through the common pathway and identify rate-limiting enzymes in the common pathway. In addition to L-phenylalanine and phenyllactate, DAH, DHS, shikimate, and shikimate 3-phosphate were observed in the culture supernatant of E. coli D2704/pKD130A. The indication that DHQ synthase is rate-limiting and DHQ dehydratase is not ratelimiting corresponds nicely with the 'H N M R analyses of E. coli aroDlpKD130A and E . coli aroElpKD130A and demonstrates that those enzymes identified to be rate-limiting or not ratelimiting are not an artifact of the host E. coli strain being scrutinized. The generality of the overall strategy to remove rate-limiting enzymes in the common pathway is further verified with the elimination of DAH accumulation upon amplification of DHQ synthase in E. coli D2704/pKD136. Observation of DHS, shikimate, and shikimate 3-phosphate in the culture supernatant of E. coli D2704/pKD136 indicated other potentially rate-limiting, common pathway enzymes. In systematic fashion, amplified expression of aroE-encoded shikimate dehydrogenase and aroL-encoded shikimate kinase eliminated both DHS and shikimate accumulation. The discovery that the rate-limiting nature of both shikimate dehydrogenase and shikimate kinase could be removed upon amplified expression of aroL came as a surprise. Feedback inhibition of shikimate dehydrogenase by shikimate had been previously reported in plants,Ig but the existence of this same feedback loop in E. coli had not previously been reported. The impact of amplified expression of aroL also indicated that an enzyme's apparent rate-limiting character may not result from inadequate expression levels or catalytic sluggishness but rather could be a consequence of feedback inhibition by the substrate of a distal, ratelimiting enzyme in the pathway. Shikimatekinase, like the aromatic aminotransferase, is part of the tyrR regulon.'"' Residual L-tyrosine from the initial growth in rich medium might contribute to transcriptional repression of shikimate kinase expression levels. Such transcriptional repression is likely to always be problematic, since even the low L-tyrosine levels resulting from de novo biosynthesis in completely unsupplemented medium are sufficient to result in accumulation of shikimate in culture supernatants of E. coli constructs.20 One approach to removing the transcriptional repression is to introduce a mutation in the tyrR locus which encodes the regulatory protein controlling expression of the various ryrR regulon enzymes. This approach was not pursued due to the adverse metabolic burden likely placed on the microbe upon derepression of all of thevarious tyrR regulon enzymes. In addition, stability of aroF-containing plasmids has been reported21 to be problematic in hosts possessing genomic mutations in tyrR. Even with the removal of DAH, DHS, and shikimate from the culture supernatant, significant improvements in the levels of L-phenylalanine and phenyllactate synthesis had still not been achieved. Attention then turned to unraveling the basis for shikimate 3-phosphate accumulation. Rate-limiting EPSP synthase (encoded by aroA) catalytic activity was an obvious possibility. However, rate-limiting chorismate synthase could also contribute, since EPSP accumulation could be converted back to shikimate 3-phosphate by EPSP synthase. Conversion of EPSP and inorganic phosphate into shikimate 3-phosphate and PEP is one commonly employed method for assay of EPSP synthase.22 Introducing plasmid-encoded aroA had a pronounced (19) (a) Balinsky, D.; Dennis, A. W.; Cleland, W. W. Biochemistry 1971,
10, 1947. (b) Dowsett, J. R.; Corbett, J. R. Biochem. J. 1971,123,23P. (c)
Lemos Silva, G. M.; Lourenco, E. J.; Neves, V. A. J . Food Biochem. 1985, 9,105. (d) Dowsett, J.R.; Middleton, B.;Corbett, J. R.;Tubbs, P.K. Biochim. Biophys. Acta 1972, 276, 344. (20) Berry, A. Genencor International, unpublished results. (21) Rood, J. I.; Sneddon, M. K.; Morrison, J. F. J. Bacteriol. 1980,144, 552. (22) Boocock, M. R.; Coggins, J. R. FEES Lett. 1983, 154, 127.
J . Am. Chem. SOC.,Vol. 115, No. 24, 1993
11587
impact on accumulation of L-phenylalanine and phenyllactate even in lieu of amplified shikimate kinase activity. Increasing the catalytic activities of both EPSP synthase and chorismate synthase was ultimately required to significantly reduce shikimate 3-phosphate levels. Plasmid-encoded aroA and aroC along with plasmid-encoded aroB and aroL ultimately led to the highest levels of carbon flow through the common pathway as evidenced by the levels of L-phenylalanine and phenyllactate synthesized by D2704/pKD136/pKAD50. Almost twice as much carbon flow is being delivered (Figure 1B) to the end of the common pathway in this construct relative to D2704/pKD130A, which expresses comparable levels of transketolase and DAHP synthase but lacks plasmid-localized genes encoding the rate-limiting pathway enzymes. Analysis of D2704 demonstrates that thecombined use of simple 'H N M R spectroscopy with the proper auxotrophic mutant can result in the straightforward identification of rate-limiting enzymes in the common pathway of a biosynthetic cascade. For the common pathway of aromatic amino acid biosynthesis, DHQ synthase, shikimate kinase, EPSP synthase, and chorismate synthase are rate-limiting enzymes. DHQ dehydratase and shikimate dehydrogenase are not rate-limiting. It is particularly important in identifying the rate-limiting enzymes to point out (Table 11) that massive overexpression of the relevant loci is not necessarily required to remove impediments to carbon flow. For instance, less than twofold increases (Table 11) in the specific activity of DHQ synthase completely compensated for the ratelimiting character of DHQ synthase. The minimum level of overexpresssion required to compensate for the rate-limiting character of DHQ synthase, shikimate kinase, EPSP synthase, and chorismate synthase remains to be established. This is an important consideration for the future, since achieving the highest possible percent conversions of D-glucose into industrial and medicinal aromatics will require that metabolic currency not be wasted on unneeded enzyme overexpression. In route to creation of such optimal microbial catalysts, the analysis methodology developed and rate-limiting enzymes identified in this study are important steps.
Experimental Section General Methods. Chorismic acid, prephenic acid, phosphoenolpyruvate, and porcine heart diaphorase were purchased from Sigma. The sodiumsaltof 3-(trimethylsilyl)propionic-2,2,3,3-d4acid was purchased from Aldrich. 14C-Shikimatewas obtained from New England Nuclear. ((Diethy1amino)ethyl)cellulose (DE-52) was obtained from Whatman. Protein assay solution was purchased from Bio-Rad. 'H NMR spectra were recorded on a Varian Gemini-200 spectrometer at 200 MHz. Bacterial Strains. Construction of BJ502aroB has been described.5b RB791Z3*(W3110 lacL8Iq) was obtained previously by this laboratory. D2704* [W3110 F- (argF-lac)delU169 (gal-bio)del (rrpde161-intc226 trplacW205) trpRGal+ tnaSaPtyrA4 (pheA)del],AB2834/pIA321,23b J P 1 6 7 2 / ~ M U 3 7 1 , ~AB2829/~KDS01,2~ )~ AB2349/pGM602,23C and AB2829/pRW5tkt/pMU377aroBwere obtained from Genencor International, Rochester, NY. AB2834z3f [tsx-352 supE42 A- aroE353 malA352 (A-)] and AB284923' (rsx-357 supE42 A- aroC355) were obtained from the E. coli Genetic Stock Center at Yale University. Culture Medium. All solutions were prepared in distilled, deionized water. LB mediumz4contained (per L) tryptone (10 g), yeast extract ( 5 g), and sodium chloride (10 g). M9 salts contained (per L) Na2HP04 (6g),KHzP04(3g),NaCl(OSg),andNH4Cl(lg). M9growthmedium contained glucose (10 g), MgS04 (0.12 g), and thiamine (1 mg) in 1 L of M9 salts. Isopropyl 8-D-thiogalactopyranoside(IPTG) (0.2 mM) was added to all cultures of strains possessing plasmids derived from pSU18 and pSU19. Chloramphenicol (20 mg/L) and ampicillin (50 mg/L) (23) (a) Frost, J. W.; Bender, J. L.; Kadonaga, J. T.;Knowles, J. R. Biochemistry 1984, 23, 4470. (b) Anton, I. A,; Coggins, J. R. Biochem. J . 1988,249,319.(c) De Feyter, R. C.; Pittard, J. J . Bacteriol. 1986,165,226. (d) Duncan, K.; Coggins, J. R. Biochem. J. 1986, 234, 49. (e) White, P.J.; Millar, G.; Coggins, J. R. Biochem. J. 1988,251,313. (f) Pittard, J.; Wallace, B. J. J. Bacteriol. 1966, 91, 1494. (24) Miller, J. H. Experimentsin MolecularGenetics;Cold Spring Harbor Laboratory: Plainview, NY, 1972.
11588 J. Am. Chem. SOC.,Vol. 115, No. 24, 1993
Dell and Frost
end fragment from pKAD43 was ligated to the prepared 1.7-kb aroC were added to all appropriate cultures. Solutions of inorganic salts, fragment, yielding pKAD5O. magnesium salts, and carbon sources were autoclaved separately and then mixed. Antibiotics, thiamine, and IPTG were sterilized through pKAD44. The 6.4-kb aroC aroA plasmid was created by isolation of 0.2-rm membranes prior to addition to the culture medium. an aroA fragment from pKAD38 followed by insertion into pKAD39. pKAD38 was digested with KpnI forming a linear DNA fragment. Culture Conditions. LB medium (1 L in a 4-L Erlenmeyer flask) Subsequent mung bean nuclease treatment followed by PsrI digestion containingIPTG and antibiotics wasinoculated with 5 mLofanovernight yielded a 2.4-kb oroA fragment. Plasmid pKAD39 was digested with culture and incubated at 37 'C for 12 h with agitation (250 rpm). Cells HindIII, the cohesive ends were removed, and the resulting DNA was were collected by centrifugation,washed three times with 3 W m L portions digested with PsrI. The resulting blunt end/PsrI pKAD39 fragment was of M9 salts, and resuspended in M9 medium (1 L) containing IPTG and ligated to the aroA fragment affording pKAD44. antibiotics. Cultures were returned to 37 Dc and 250 rpm agitation. pKAM1. The 5-kb aroCaroL plasmid was constructed by isolation NMR Analysis of Culture Supernatant. An aliquot (25 mL) of culture of aroCfrom pKAD39 followed by its ligation into pKAD31. pKAD39 supernatant was takenattheindicated timeintervalandcellswereremoved was digested with KpnI, blunt ended, and digested with SalI to liberate by centrifugation. The supernatant was neutralized by the addition of the aroCgene. pKAD31 was linearized withBamH1, treated with mung 2 N NaOH, concentrated to dryness several times from D20, and bean nuclease, and digested with SalI. The resulting blunt ended/SalI redissolved in D2O containing a known concentration of the sodium salt pKAD31 fragment was ligated to aroC. acid (TSP). Concentrations of of 3-(trimethylsilyl)propionic-2,2,3,3-d~ pKAD42. The 10-kbaroFaroBplasmid wasconstructedfrompKD136 cellular metabolites in the supernatant were determined by comparison by removal of the 5-kb rkr insert as a BamHI fragment followed by of the integrals of known metabolite resonances and the resonance rccircularization. correspondingto TSP in the 'H NMR. Cultures were grown in triplicate pKAD46A. The 7-kb serCaroA plasmid was constructed by isolating to establish mean values and standard deviations. a 4.7-kb PsrI fragment containing the serCaroA operon from ~ K D 5 0 1 2 ~ Genetic Manipulstio~. Recombinant DNA manipulations were and inserting it in the PsrI site of pSU19. performed as outlined in Sambrook et al.25 T4 DNA ligase was used for pKADS2. Construction of the 4.2-kb pheA plasmid was achieved by all ligations, and blunt ends were formed by removing protruding ends isolatingpheA from as an EcoRI/Bspl286I fragmentZBfollowed of DNA with mung bean nuclease. Construction of pKD130As' and by insertion into pSU 18. pKD136" havebeendescribedelsewhere. Thevectors pSU18 andpSU19 Preparation of DAHP, DHS, ShiLimrte 3-Phosphrte, and EPSP. are medium copy number vectors derived from the plasmids pSU2718/ Strains for metabolite accumulations were grown in LB medium and pSU27 19.26 Plasmids pSU18 and pSU19 possess chloramphenicol resuspended in M9 medium as described under culture conditions. resistance, a lac promoter, and a p15A origin of replication and differ Metabolites were isolated from the culture supernatant upon removal of in the orientation of the pUC18 multiple cloning site with respect to the cells. lac promoter. Plasmids carrying loci encoding rate-limiting, common 3-Deoxy-~-arabino-heptulosonate 7-phosphate was synthesized from pathway enzymes were constructed using E. coli DNA in the vectors according to methyl (methyl 3-deoxy-~-arabino-heptulopyranosid)onate pSU18 and pSU19 and are shown in Table I. theprocedureof Frost and K n o w l ~ 8 .DAH ~ ~ wassynthesiztdasdescribed pKD28. A 1.6-kb BamHI/HindIII fragment containing the aroE gene expressed from a roc promoter was isolated from ~ I A 3 2 and 1 ~ ~ ~ previously using E. coli BJ502~roB/pKD130A.~DAH was purified and subsequently converted to methyl (methyl 3-deoxy-~arabinoligated into the vector pSU18. pKAD31. A 1.0-kb BamHIIKpnI aroL fragment from ~ M U 3 7 1 ~ ~heptu1opyranosid)onateas described by R ~ i m e r . ~ ~ 3-Dehydroshikimatewas isolated from the culture supernatant (1 L) was isolated and ligated into pSU19. E. coli AB2834uroE/pKD136." Following removal of the cells, the of pKAD34. This 4.9-kb aroE aroL plasmid was constructed by supernatant (24 mmol of DHS) was adjusted to pH 2.5 by addition of manipulationof the flanking restrictionsites of aroE from pKD28 followed 6 N HCl. After continuous extraction with ethyl acetate, the organic by ligation into pKAD31. pKD28 was linearized with BamHI, and the fraction was dried over MgSO4 and concentrated. Subsequent rccrys5' protruding ends of DNA were removed. XbaI linkers were ligated to tallization of the brown oil from ethyl acetate afforded 1.6 g of DHS the resulting blunt ends, and the plasmid was recircularized forming (40% yield). pKAD32. pKAD32 was digested with HindIII and blunt ended, BamHI Shikimate 3-phosphate was isolated from the culture supernatant of linkers were attached, and the plasmid was recircularized forming AB2829/pRW5rkr/pMU377oroB. Host strain AB2829 accumulata pKAD33. The aroE gene, now localized on a 1.6-kb XbaI/BamHI shikimate 3-phosphatedue to a mutation in the gene which encodesEPSP fragment, was isolated from pKAD33 and ligated into pKAD3 1 to form synthase. Plasmid pRW5rkPb contains a feedback-resistant u r d gene pKAD34. and rkr. Plasmid pMU377aroB, a pBR322m derivative, encodes aroL pKAD38. This 4.7-kb plasmid was created by ligation of a 2.4-kb and aroB. Two liters of cell-free culture supernatant (3.6 mmol of KpnIIPstI fragment encoding aroA derived from ~ K D 5 0 1 behind 2 ~ ~ the shikimate 3-phosphate)were passed throughacolumn (500mL) ofDowex lac promoter of pSU18. The absence of a native promoter on the aroA 50 (H+ form) at 4 OC. The column was washed with 1.5 L of water, and fragment necessitates its placement behind a promoter for expression. the resulting eluent was adjusted to pH 8.0 by addition of 5 M LiOH. pKAD43. This 5.7-kb aroA aroL plasmid was created by isolation of Water was removed from the sample in vacuo. Methanol ( 5 0 0 mL) was an aroA fragment from pKAD38 followed by its ligation into pKAD3 1. added to the resulting solid, and the mixture was stirred at 4 O C for 1.5 pKAD38 was linearized with KpnI, and the resulting 3' ends of DNA h. After removal of the precipitate by filtration, the sample was were removed. Subsequent digestion with PsrI yielded a 2.4-kb blunt concentrated to dryness. The residue was dissolved in water (50 mL), end/PstI aroA fragment. pKAD3 1was digestedwith Hind111 and treated the pH of the resulting solution was adjusted to 7.0 by addition of 1 M with mung bean nuclease. Digestion with PsrI afforded a 3.3-kb blunt LiOH, and a white precipitate was removed by filtration and discarded. end/PsrI linearized vector suitable for ligation with the prepared aroA The solution was loaded onto a column of DE-52 ( 5 cm X 19.5 cm) at fragment. 4 'C. The column was washed with 1 column volume of water and pKAD39. This 4.0-kb aroC plasmid was constructed by isolating an subsequently eluted with a linear gradient (1.5 L + 1.5 L,0-400 mM) aroC fragment from pGM602230followed by its insertion into pSU19. of triethylammonium bicarbonate, pH 7.2. Column fractions (25 mL) pGM602 was digested with ClaI and treated with mung bean nuclease. were analyzed for total phosphorous and for inorganic phosphate.'! Digestion with SalI yielded a 1.7-kb SalI/blunt end fragment. pSU19 Fractions that contained shikimate 3-phosphatebut were freeof inorganic was digested simultaneously with Sun and SmaI and ligated to the phosphate were pooled and concentrated under reduced pressure to 100 prepared aroC fragment. mL. Buffer was removed by azeotropicdistillationwith 2-propanol (three times). After concentrating the sample, the resulting white powder was pKADS0. The 7.4-kb oroA aroC aroL plasmid was constructed by dissolved in water and passed down a column (100 mL) of Dowex 50 inserting the arocfragment from pKAD39 intopKAD43. pKAD39 was linearized with KpnI, and the 3' protruding ends of DNA were removed. (27) Zurawski. G.; Brown, K.;Killingly, _ _ D.; Yanofsb, C. h o c . Nut/.Acad. A 1.7-kb SalI/blunt end aroC fragment was isolated .upon digestion of Sci. USA. 1978, 75, 4211. the DNA withSuI1. pKAD43 wasdigested withBomHIand blunt ended, (28) Stewart, 1.;Wilson, D. B.;Ganem, B. Tetrahedron 1991,47,2573. and the resulting DNA was digested with SalI. The 5.7-kb SalI/blunt (29) Reimer, L. M.; Conley, D. L.; Pompliano, D. L.; Frost, J. W. J. Am. Chem.'Soc. 1986, 108, 8010.(30) Balbas, P.; Soberon, X.;Merino, E.; Zurita, M.;Lomeli, H.; Valle, (25) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A F.; Flores. N.:Bolivar. F. Gene 1986, 50, 3. Laboratory Manual; Cold Spring Harbor Laboratory: Plainview. NY,1990. (31) Ames, B. N. keth. Enzymol; 1%. 8, 115. (26) Martinez, E.; Bartolome, B.; de la Cruz, F. Gene 1988, 68, 159.
Biocatalytic Synthesis of Aromatics from D-Glucose (Na+ form). The sample was eluted with three column volumes of water and concentrated to dryness to afford 0.21 g of the trisodium shikimate 3-phosphate (18% yield). Spectroscopic characterization of the purified shikimate 3-phosphate correctly correlated with literature data.32 5-Enolpyruvoylshikimate 3-phosphate was synthesized enzymatically as previously described33from phosphoenolpyruvateand sodium shikimate 3-phosphate. EPSP synthase was partially purified from DH5a/ pKAD46A using the procedure of Levin and Sprins0n.~3~ Enzyme Assays. Cells were grown in LB media for 12 h as described under culture conditions. Cells were isolated by centrifugation and disrupted by two passages through a French press at 16 000 psi. Cellular debris was removed by centrifugation at 48 OOOg for 20 min. Protein was quantified using the Bradford dye-binding procedure.34 A standard curve was prepared with bovine serum albumin. Chorismate mutase was assayed” using the method of Cotton et al. Prephenic dehydratase was assayed35 by following the formation of phenylpyruvic acid from prephenic acid at 320 nm. DHQ synthase activity was measured by monitoring the disappearance of DAHP over time23a using the thiobarbituric acid a ~ s a ytoquantify 3~ DAHP. Shikimatekinase was assayed by monitoring the formation of I4C-shikimate 3-phosphate from I4C-shikimate using the method of De Feyter.” EPSP synthase wasassayed in the reversedirection as describedby Boocockand Coggins.2z Chorismate synthase was assayed38 by monitoring the formation of chorismate through its subsequent conversion to phenylpyruvate at 320 (32) Bartlett, P. A,; McQuaid, L. A. J. Am. Chem. SOC.1984,106,7854. (33) (a) Knowles, P. F.; Levin, J. G.; Sprinson, D. B. Meth. Enzymol. 1971, 17A, 360. (b) Levin, J. G.; Sprinson, D. B. J. Eiol. Chem. 1964, 239, 1142. (34) Bradford, M. M. Anal. Biochem.1976, 72, 248. (35) Cotton, R. G. H.; Gibson, F. Meth. Enzymol. 1971, 17A, 564. (36) Gollub, E.; Zalkin, H.; Sprinson, D. B. Meth. Enzymol. 1971, 17A, 349. (37) De Feyter, R. Meth. Enzymol. 1987, 142, 355.
J. Am. Chem. SOC.,Vol. 115, No. 24, 1993
11589
nm using chorismate mutase and prephenate dehydratase as coupling enzymes. FADHz required for activity was generated in situ from FAD, NADH, and porcine heart diaphorase as described.38 A crude extract of the coupling enzymes was prepared from AB2849/pKAD52. Inhibition of ShikimateDehydrogenase. Shikimatedehydrogenase was purified39a from AB2834/pIA321 and stored at -70 OC at a final concentration of 1600 units/mL in 50% glycerol, 50 mM Tris-HCl, pH 7.5, 1 mM benzamidine, 0.4 mM dithiothreitol, and 50 mM KC1. For the inhibition studies, the conversion of dehydroshikimate to shikimate with the subsequent oxidation of NADPH was monitored39bat 340 nm and 24 ‘C. The assay contained (final volume 1 mL) 0.1 M KH2P04, pH 7.4, 0.21 mM NADPH, DHS, shikimate, and 0.01 mL of diluted enzyme. The Ki for shikimate was determined under the following conditions: NADPH (Km0.03 mM) was held constant at 0.21 mM; DHS concentrations (Km0.072mM) were 0.033,0.050,0.067,0.10, and 0.20 mM; shikimate concentrations were 0, 0.13, 0.25, 0.38, and 0.50 mM. Enzymesampleswerediluted 1:4000 directly from the frozen stock in 0.05 M Na2HP04, pH 7.4, 0.1 mM dithiothreitol, and 0.1 mg/mL acetylated bovine serum albumin. The Ki for shikimate was obtained from a replot of the slope of a Lineweaver-Burp plot (Figure 3) versus inhibitor concentration. Points at high inhibitor concentrations and low substrateconcentrationsin the Lineweaver-Burkplot were found to deviate from linearity and were removed.
Acknowledgment. This research was supported by a contract from Genencor International, Inc., and a grant from the National Science Foundation. We are indebted to Karen M. Draths for many helpful discussions and comments. (38) Morell, H.; Clark, M.J.; Knowles, P. F.;Sprinson, D. B. J. Bioi
Chem. 1967, 242, 82.
(39) (a) Chaudhuri, S.;Anton, I. A.; Coggins, J. R. Meth. Enzymol. 1987, 142,315. (b) Coggins, J. R.; Boocock, M. R.; Chaudhuri, S.; Lambert, J. M.; Lumsden, J.; Nimmo, G. A.; Smith, D. D. S.Meth. Enzymol. 1987,142,325. (40) Lineweaver, H.; Burk, D. J . Am. Chem. Soc. 1934, 56,658.
